performance tuning for 1.0

[simonmichael]

https://groups.google.com/forum/#!topic/hledger/sA0lA7wLUFo

After posting those ideas, I remembered that actually I'm trying to get to 1.0, and we have enough features for that. What we don't have enough of is speed; we are generally slower than Ledger, at least (and especially) when files get large. At least, that's my understanding and common wisdom, though, some of the benchmark results like http://hledger.org/profs/201409131354.bench tell a different story. I suppose that's an isolated case due to an incorrect Ledger build. More clarity needed here. Regardless, I know hledger isn't as fast as I'd like if I run it on large multi-year journals. And 0.24's more powerful account aliases are not performing well (https://github.com/simonmichael/hledger/issues/244).

I think it's possible for a Haskell implementation to match or exceed Ledger's performance. We should try to do that. And then release 1.0.